AI Offers Guidance

Whether automating routine tasks, real-time analyses or precise demand and price development forecasts, buyers can save a lot of time and money, reduce errors and risks, improve decision-making and increase their productivity by using AI. Nevertheless, many companies still shy away from introducing AI and, according to the Federal Statistical Office, cite a lack of knowledge (72%) as the main reason, followed by incompatibility with existing systems (54%) and the availability or quality of data (53%). As far as the lack of in-house expertise is concerned, concerns can be overcome with the help of an AI roadmap that is tailored to the specific tasks of individual procurement organizations and includes various milestones.

Kick-off with an analysis of the current situation

This begins with the analysis of existing operational (request-to-order) and strategic (source-to-contract) purchasing processes for the possibility of optimizing them through the use of intelligent tools. apsolut operates a constantly growing library with currently around 80 AI use cases: from the automation of recurring manual tasks such as order processing, supplier qualification and simple negotiations to AI-supported data analyses for risk management, contract management and supplier evaluations. There are also use cases for supplier development and complex negotiation strategies. Depending on the use case, different types of AI are available, including machine learning (ML), deep learning (DL) or large language models (LLMs) such as GPT-4.

Weighing up costs and benefits

Whether an AI implementation for certain process steps actually pays for itself becomes clear through a cost-benefit analysis based on predefined goals and KPIs. One example is the introduction of LLM negotiation bots for tail-spend orders, i.e. for non-strategic transactions with a low value or for infrequent purchases by a company whose total volume can accumulate considerably.

In order to achieve the return on investment (ROI), the LLM negotiation bids should allow for cost savings of at least three percent in tail spend. To keep the effort as low as possible, the cost-benefit analyses are carried out as part of a proof of concept (POC). In the case of a negotiation bot, the provider provides the purchasing organization with the corresponding app, which - without having to be integrated into existing systems - can be used for selected sample requirements in a short space of time. If the POC is successful and proves the expected cost savings, the procurement-wide implementation of the LLM negotiation bot is an option.

AI tools with SAP integration

The technical hurdles cited by companies - the incompatibility of systems and problems with data availability - must be overcome. SAP-based purchasing organizations are therefore well advised to choose intelligent solutions that are based on the Business Technology Platform (SAP BTP). This enables the AI tools to communicate seamlessly with the existing SAP purchasing solutions and access all the required data in real time. Examples of such tools that can be easily integrated into SAP include the apsolut Negotiation Bot for automated supplier negotiations and the apsolut Procurement Bot, which supports the ordering of indirect materials. Together with Lhotse, an expert in AI-based user guidance in procurement, apsolut has also developed the "Copilot for SAP Ariba", which reduces the average processing time for free text purchase requisitions from twelve to five days.

Even if the topic of AI in procurement is still in its infancy in many places, the continuous technological progress means that intelligent solutions are also becoming increasingly important in this business area. New AI application areas are emerging and producing increasingly sophisticated AI tools that offer growing benefits and relieve buyers of tedious routine tasks. They can concentrate more on strategic and value-adding tasks.

SAP-based purchasing organizations should therefore put aside any reservations they may have about AI as quickly as possible - and set the most important course for the future with a clear AI roadmap and the right AI tools.

To the partner entry: